Upload processor
Browse files- vocab.json +112 -1
vocab.json
CHANGED
|
@@ -1 +1,112 @@
|
|
| 1 |
-
{
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"%": 6,
|
| 3 |
+
"0": 77,
|
| 4 |
+
"1": 44,
|
| 5 |
+
"2": 21,
|
| 6 |
+
"3": 9,
|
| 7 |
+
"4": 60,
|
| 8 |
+
"5": 35,
|
| 9 |
+
"6": 69,
|
| 10 |
+
"7": 64,
|
| 11 |
+
"8": 84,
|
| 12 |
+
"9": 20,
|
| 13 |
+
"[PAD]": 109,
|
| 14 |
+
"[UNK]": 108,
|
| 15 |
+
"a": 27,
|
| 16 |
+
"b": 17,
|
| 17 |
+
"c": 75,
|
| 18 |
+
"d": 10,
|
| 19 |
+
"e": 4,
|
| 20 |
+
"f": 47,
|
| 21 |
+
"g": 37,
|
| 22 |
+
"h": 2,
|
| 23 |
+
"i": 97,
|
| 24 |
+
"k": 1,
|
| 25 |
+
"l": 29,
|
| 26 |
+
"m": 59,
|
| 27 |
+
"n": 3,
|
| 28 |
+
"o": 67,
|
| 29 |
+
"p": 26,
|
| 30 |
+
"q": 71,
|
| 31 |
+
"r": 19,
|
| 32 |
+
"s": 36,
|
| 33 |
+
"t": 46,
|
| 34 |
+
"u": 43,
|
| 35 |
+
"v": 98,
|
| 36 |
+
"w": 50,
|
| 37 |
+
"x": 79,
|
| 38 |
+
"y": 106,
|
| 39 |
+
"z": 76,
|
| 40 |
+
"|": 32,
|
| 41 |
+
"à": 24,
|
| 42 |
+
"á": 8,
|
| 43 |
+
"â": 103,
|
| 44 |
+
"ã": 80,
|
| 45 |
+
"è": 107,
|
| 46 |
+
"é": 45,
|
| 47 |
+
"ê": 30,
|
| 48 |
+
"ì": 70,
|
| 49 |
+
"í": 56,
|
| 50 |
+
"ò": 90,
|
| 51 |
+
"ó": 102,
|
| 52 |
+
"ô": 57,
|
| 53 |
+
"õ": 42,
|
| 54 |
+
"ù": 85,
|
| 55 |
+
"ú": 100,
|
| 56 |
+
"ý": 54,
|
| 57 |
+
"ă": 72,
|
| 58 |
+
"đ": 52,
|
| 59 |
+
"ĩ": 99,
|
| 60 |
+
"ũ": 101,
|
| 61 |
+
"ơ": 40,
|
| 62 |
+
"ư": 41,
|
| 63 |
+
"̀": 12,
|
| 64 |
+
"́": 49,
|
| 65 |
+
"̣": 51,
|
| 66 |
+
"у": 33,
|
| 67 |
+
"ạ": 89,
|
| 68 |
+
"ả": 93,
|
| 69 |
+
"ấ": 82,
|
| 70 |
+
"ầ": 95,
|
| 71 |
+
"ẩ": 91,
|
| 72 |
+
"ẫ": 5,
|
| 73 |
+
"ậ": 68,
|
| 74 |
+
"ắ": 31,
|
| 75 |
+
"ằ": 7,
|
| 76 |
+
"ẳ": 62,
|
| 77 |
+
"ẵ": 81,
|
| 78 |
+
"ặ": 73,
|
| 79 |
+
"ẹ": 34,
|
| 80 |
+
"ẻ": 78,
|
| 81 |
+
"ẽ": 28,
|
| 82 |
+
"ế": 105,
|
| 83 |
+
"ề": 53,
|
| 84 |
+
"ể": 92,
|
| 85 |
+
"ễ": 14,
|
| 86 |
+
"ệ": 39,
|
| 87 |
+
"ỉ": 88,
|
| 88 |
+
"ị": 83,
|
| 89 |
+
"ọ": 48,
|
| 90 |
+
"ỏ": 0,
|
| 91 |
+
"ố": 66,
|
| 92 |
+
"ồ": 25,
|
| 93 |
+
"ổ": 38,
|
| 94 |
+
"ỗ": 65,
|
| 95 |
+
"ộ": 86,
|
| 96 |
+
"ớ": 16,
|
| 97 |
+
"ờ": 23,
|
| 98 |
+
"ở": 11,
|
| 99 |
+
"ỡ": 96,
|
| 100 |
+
"ợ": 58,
|
| 101 |
+
"ụ": 63,
|
| 102 |
+
"ủ": 13,
|
| 103 |
+
"ứ": 15,
|
| 104 |
+
"ừ": 22,
|
| 105 |
+
"ử": 104,
|
| 106 |
+
"ữ": 61,
|
| 107 |
+
"ự": 94,
|
| 108 |
+
"ỳ": 74,
|
| 109 |
+
"ỵ": 87,
|
| 110 |
+
"ỷ": 18,
|
| 111 |
+
"ỹ": 55
|
| 112 |
+
}
|